[Rd] rbind.data.frame drops attributes for factor variables

[Erik Iverson]

Dear R-devel -

I noticed that when I rbind two data.frames together, factor variables 
lose their attributes in the resulting data.frame while numeric 
variables do not.

As an example, create two data.frames, t1 and t2, with two variables 
each.  Give each variable an attribute called "label", and then perform 
the rbind and look at the resulting structure.

#### EXAMPLE R CODE #####
t1 <- data.frame(subject = 1:4, trt = factor(c("A","B","B","A")))
attr(t1$trt, "label") <- "Trt Label"
attr(t1$subject, "label") <- "Subject Label"
str(t1)

t2 <- data.frame(subject = 5:8, trt = factor(c("A","A","A","A")))
attr(t2$trt, "label") <- "Trt Label"
attr(t2$subject, "label") <- "Subject Label"
str(t2)

str(rbind(t1, t2))

#### END EXAMPLE R CODE #####

The output of the last line of code is:

'data.frame':	8 obs. of  2 variables:
  $ subject: atomic  1 2 3 4 5 6 7 8
   ..- attr(*, "label")= chr "Subject Label"
  $ trt    : Factor w/ 2 levels "A","B": 1 2 2 1 1 1 1 1

I do not see this documented anywhere in ?rbind, and do not know if it 
is intended.  It looks like the factor loses its attributes in 
rbind.data.frame due to a call to as.vector.  Of course, as.vector is 
documented to drop the attributes of atomic vector types.  I do not know 
if this qualifies as a bug or intended since it is not stated in ?rbind 
what will happen. sessionInfo() is below.

Best,
Erik Iverson

sessionInfo()
R version 2.7.0 (2008-04-22)
i686-pc-linux-gnu

locale:
LC_CTYPE=en_US.UTF-8;LC_NUMERIC=C;LC_TIME=en_US.UTF-8;LC_COLLATE=en_US.UTF-8;LC_MONETARY=C;LC_MESSAGES=en_US.UTF-8;LC_PAPER=en_US.UTF-8;LC_NAME=C;LC_ADDRESS=C;LC_TELEPHONE=C;LC_MEASUREMENT=en_US.UTF-8;LC_IDENTIFICATION=C

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base